It is often desirable that a controller and the power switching means it controls be combined in a two terminal device intended for connection in series with a source of electrical energy and the load being switched. A timer control for a lighting load, for example, so designed could be substituted for the ordinary wall switch which normally controls the light, even if both sides of the power line are not available in the switch box.
Such an arrangement, however, requires that the power supply for the controller receive its energy through the load during both states of the power switch; that is, when it is open or closed. An additional practical requirement of such a system is that the energy consumed by the controller be very small in comparison to the energy consumed by the load. This latter requirement is easily met by modern controllers, even for controlling loads as small as a few watts. Thus, the minute amount of power required by the controller can be supplied through the load when the power switch is open, the current then passing through the load being so small in comparison to the full load current that the load, for all practical purposes, is de-energized. When the power switch is closed, however, energy for the controller must be obtained in a different manner.
The prior art teaches two methods for deriving a minute amount of power for the controller when such a power switch is closed, one method being shown in U.S. Pat. No. 3,491,249 to Rabinow, and the other by U.S. Pat. No. 3,940,660 to Edwards. In the method shown in U.S. Pat. No. 3,491,249 to Rabinow a constant low impedance is placed in series with the load, so that the voltage drop across this impedance is only a small fraction of the voltage of the power source. Rabinow's controller is an electric clock mechanism. He achieves the low impedance simply by switching the load from a high impedance low current clock coil to a low impedance high current winding on the clock coil for which the full-load current produces only a small voltage drop but still supplies sufficient power at the relatively high load current to keep the clock running.
In U.S. Pat. No. 3,940,660 to Edwards (and also in later patents of others) there is disclosed a time division method of sharing power between the load and the controller when the power switch is closed. This method involves delivering energy to the controller power supply in the extremely short intervals during which a threshold type power switch like a triac is non-conductive at the beginning of each half cycle of the applied AC voltage.
If the method of Rabinow were to be applied to the power supply for an electronic controller rather than a clock mechanism controller, a transformer could be substituted for the clock motor coil. However, such a transformer would be bulky and costly in comparison to the method of this invention. Other series-connected low impedance devices could be used instead of a transformer, however, but these also have their drawbacks. The series impedance could, for instance, be a small saturable core inductor but this would also be bulky and expensive in comparison to this invention. Back-to-back diodes could also be used. These would have the advantage of causing a relatively constant voltage drop. However, to obtain voltage high enough to power a typical electronic circuit might require stacking two or more pairs of back-to-back diodes or the use of back-to-back zener diodes. Such a system would be much less bulky than a transformer or saturable inductor. However, it has the serious drawback that the power consumed by the diodes is dependent upon the load current. If a wide range of loads is to be switched, such as lamp loads from 25 to 600 watts, the current range for a 120 VAC supply would be from about 0.2 to 5.0 amperes. If a 3.0 volt drop across the diodes were to be required for the electronic controller power supply, the power dissipation of the diodes could rise to 15 watts. This would preclude the design of a controller of small enough size to operate in a wall switch box because of excessive temperature rise. The situation would be even worse if the series impedance were purely resistive.
While the just described problems are avoided by the energy time sharing system described in the Edwards patent, this system has significant disadvantages. Thus, this time sharing system supplies the power to the controller power supply through a resistance (R1 in FIG. 2 of the Edwards patent) which is continuously connected to the AC power source through the load. When the load is switched "off", the power supply consisting of diodes (D1, D2, Z1, Z2) and capacitors (C1 and C2) and resistance R1 are in series with the AC power source and the load. Since the load is a low impedance compared to that of the resistance R1, resistance R1 must have a wattage rating nearly equivalent to the power it would dissipate if it were connected directly across the AC power source. Therefore, the value of resistance R1 should be made as high as possible when the load is switched "off", consistant with the small amount of power actually required by the controller circuitry. If the full AC power source voltage were always available, the resistance R1 could be made so high as to reduce its power dissipation to well under one watt. However, when the load is to be switched "on", the voltage for the control power supply is available only during those brief time-share instances when the load switch (in this case a triac) is non-conducting.
If the triac switch is to be non-conducting for a portion of each cycle or half cycle of the AC power source, then the triggering of the triac must be delayed from the moment of each zero crossing until the supply voltage has risen to a value sufficient to supply the required energy through resistance R1. However, it is very desirable that the voltage reached by the supply before the triac is switched on be as low as possible so as to deliver the maximum amount of power to the load, and even more important, to prevent the production of radio interference noise caused by the rapid switching characteristics of the triac. Studies have shown that switching incandescent lamps in a 120 volt AC circuit by means of a triac or the like requires the use of a filter network to suppress radio interference, unless the triac is switched very close to the "zero crossing" of the applied AC voltage, that is, prior to the voltage having increased (positively or negatively from zero) to more than about 5 volts. To maintain switching near zero crossing and still supply enough power through resistance R1 for the control power supply requires that resistance R1 should have as low a resistance as possible consistant with supplying the energy required by the control circuitry. Thus, there are contradictory requirements for the value of resistance R1 between times when the load is to be switched "off", where it is desirable to have resistance R1 a very high value, and when the load is to be switched "on", where it is desirable to have resistance R1 a very low or even zero value. The compromise required between these two desirable values of resistance R1 has been found to preclude the use of a low wattage resistor for resistance R1 and thus the time sharing system disclosed in the Edwards patent has the same excessive power dissipation problem of the series impedance method described previously. If a compromise with delivering full power to the load is made so that resistance R1 can have a substantially high value, then the supply voltage must rise to a value in excess of 50 volts and perhaps to as high as 100 volts, before the triac is triggered. In this case, however, a noise filter will be required to suppress radio interference and the cost and bulk will be considerably increased.
A second disadvantage of the time sharing system described is the manner in which both positive and negative gate current for triggering the triac is obtained. (It is most desirable to trigger a triac with a positive gate current for one half cycle and a negative gate current on the opposite half-cycle, since this requires the lowest value of gate current to assure triggering and the performance is thus most easily guaranteed by the manufacturer). In the system disclosed in the Edwards patent, to provide for dual polarity gate current, two control power supplies of opposite polarity must be supplied, thereby doubling the cost of the power supply. Also, since the gate current must be switched by logic circuits, discrete transistors or other control devices are required to switch the gate current, such as an NPN transistor (Q1) and PNP transistor (Q2). Since these transistors are reverse biased between emitter and base for one half cycle by the peak source voltage, they must be protected by the addition of diodes (D3 and D4). The need to use all these components considerably increases the cost and bulk of the controller as compared to a circuit like that of the present invention. However, the most serious disadvantage is that the value of resistance R1 required to prevent excessive power dissipation when the load is switched off requires delaying the triac trigger for the switched on state until the AC power source reaches a relatively high value, thus requiring the addition of a filter to eliminate radio interference noise.
The present invention avoids these disadvantages and makes possible the design of an extremely compact, low cost controller for a series connected load in which the total power dissipation is essentially that of the triac switch and which also provides near zero crossing switching in a manner that a radio interference filter is not needed.